In battery electric vehicles (BEV), energy originates in the battery and is transmitted to the wheels through a series of energy conversion processes involving the inverter and motor. Therefore, understanding the energy conversion mechanisms in both the inverter and motor is essential for accurately modeling energy consumption. However, in previous studies, real-world driving data are often limited, making it challenging to fully analyze the complex and nonlinear relationships within each conversion component. In this study, we collected input–output data from the inverters and motors of fifty-four BEVs, measured repeatedly over time. The data revealed a piecewise nonlinear relationship between input and output, prompting us to partition the models by different phases: propulsion, regeneration, and battery status. For each phase, we applied linear mixed-effects models to account for the hierarchical structure of the data, estimating coefficients separately for the inverter and motor using a randomly selected 75% of the dataset. Through this component-level modeling approach, the models not only capture component-level random-effect parameters but also effectively model the nonlinear energy conversion characteristics at the component level. The two models were then integrated to estimate the total driving energy consumption of the BEVs, and the results were validated against actual observations using the total driving energy from the remaining 25% of the dataset. Model performance was evaluated using the Total Consumption Estimation Rate (TCER) and Mean Absolute Percentage Error (MAPE). The proposed model achieved at least 95.27% in TCER and 86.34% in MAPE, outperforming existing approaches with a 20% higher TCER and an MAPE approximately ten times lower on average. The comparison demonstrated that our model accurately estimates driving energy consumption, as it effectively captured the heterogeneous and nonlinear relationships between input and output energy for each component.
扫码关注我们
求助内容:
应助结果提醒方式:
